Artificial Intelligence has been used to produce all manner of creative outputs – from fiction to works of art. However, it’s taken a long time for it to be tasked with a fairly obvious job: writing code.
Computers Coding Each Other
Computers run on code, and programming languages have exactly the kind of logical, structured syntax that make them comprehensible to computers. It seems that 2022 is the year when AI-assisted coding hits the mainstream. While there’s no danger (yet) of programmers being replaced by machines, it seems ever-more plausible that computers can put up the low-level grunt work involved in code, leaving the humans to focus on critical and creative thinking.
Having been glimpsed in preview mode, GitHub Copilot will become generally available for developers in summer 2022. Trained on billions of lines of publicly available source code and natural language text, the tool makes real-time suggestions for your next line of code. Think predictive text on steroids.
The model that powers GitHub Copilot is Codex, a model that reads natural language text and can interpret it. This opens up possibilities for allowing somebody with zero coding skills to simply type in their requirements (“Blue sign-up button in the middle of the page”), and the model turns that into a coded-up prototype.
There are plenty of other tech companies moving into this space. Tabnine has picked up the patterns from millions of code repositories, but will also update its suggestions based on your own code – so as your dev teams ship more features, its suggestions will get more accurate.
“By automating mundane coding tasks, AI coding assistants can save developers time and money, while helping to reduce errors and improve code quality.”
Another company, Kite, is focused on speed and efficiency. It estimates that using its tool results in 47% fewer keystrokes, as the model picks up on the repetitious code blocks and provides them for you.
Your next, best coding hire might just be a cyborg.